taiji-topo-file-downloader_skill

Overview

This skill automates locating the node sidebar on the Taiji web platform, opens the configuration file panel, and downloads selected topology or model files to a timestamped local directory. It emphasizes page pre-checks and layered UI locating to avoid misclicks and failures when the page layout or classes change. The workflow includes verification, sidebar detection, file-list navigation, download initiation, and safe local renaming and placement.

How this skill works

The skill first asserts the browser tab is on a.taiji.woa.com and that the workflow canvas or sidebar is visible. It then locates the node parameter sidebar by layered strategies: semantic text anchors, structural container patterns (Ant Design), and style-class fallbacks with text revalidation. Inside the sidebar it opens the file manager modal, finds the target file row, clicks the download action, and finally moves and renames the temporary Chrome download into a {topo}_{YYYYMMDD_HHMMSS} directory. Each step includes assertions and recovery paths for robustness.

When to use it

• Automating downloads of topology config or model files from Taiji workflow pages
• Batch backup or archive tasks that must reliably find files despite UI changes
• Headless/browser-automated pipelines that need deterministic download completion checks
• Tasks requiring safe renaming and directory organization of Chrome temporary downloads
• Scenarios where page load timing or element drift has caused flaky UI automation

Best practices

• Always validate current tab URL contains a.taiji.woa.com before interacting
• Locate the sidebar using semantic text anchors first, structure containers second, and class-based matching only as a fallback
• After each critical click assert the expected UI (sidebar visible, modal present, table rows found) before proceeding
• Match file names by full name first; fallback to includes only when necessary
• Use the timestamped topo_name_{YYYYMMDD_HHMMSS} output directory to prevent overwrites and support traceability
• When using OpenClaw, prefer snapshot-based refs and browser.act calls over coordinate clicks

Example use cases

• Automated nightly backup of all topology configuration files from multiple Taiji topologies
• CI pipeline step that pulls a model file from a Taiji node and packages it into a build artifact
• Support script for operators: open a Taiji workflow, verify node sidebar, and download selected files into a dated archive folder
• Ad-hoc recovery: find a missing model file on the Taiji page and save it locally with a safe name
• Browser automation job that must validate download completion by checking Chrome temp files and renaming them